Countries are increasingly interested in spacecraft surveillance and recognition which play an important role in on-orbit maintenance, space docking, and other applications. Traditional detection methods, including radar, have many restrictions, such as excessive costs and energy supply problems. For many on-orbit servicing spacecraft, image recognition is a simple but relatively accurate method for obtaining sufficient position and direction information to offer services. However, to the best of our knowledge, few practical machine-learning models focusing on the recognition of spacecraft feature components have been reported. In addition, it is difficult to find substantial on-orbit images with which to train or evaluate such a model. In this study, we first created a new dataset containing numerous artificial images of on-orbit spacecraft with labeled components. Our base images were derived from 3D Max and STK software. These images include many types of satellites and satellite postures. Considering real-world illumination conditions and imperfect camera observations, we developed a degradation algorithm that enabled us to produce thousands of artificial images of spacecraft. The feature components of the spacecraft in all images were labeled manually. We discovered that direct utilization of the DeepLab V3+ model leads to poor edge recognition. Poorly defined edges provide imprecise position or direction information and degrade the performance of on-orbit services. Thus, the edge information of the target was taken as a supervisory guide, and was used to develop the proposed Edge Auxiliary Supervision DeepLab Network (EASDN). The main idea of EASDN is to provide a new edge auxiliary loss by calculating the L2 loss between the predicted edge masks and ground-truth edge masks during training. Our extensive experiments demonstrate that our network can perform well both on our benchmark and on real on-orbit spacecraft images from the Internet. Furthermore, the device usage and processing time meet the demands of engineering applications.

This paper presents a method for spacecraft to achieve the task of fast maneuvering and fast stabilization. To realize this task, a new type of vibration isolation platform whose actuators are based on magnetic suspension techniques, and an attitude controller to suit the spacecraft with this new vibration isolation platform are presented. High frequency vibrations would reduce the stability of the attitude control, and low frequency vibrations would reduce the maneuvering time of the attitude control. The vibration isolation platform presented in this paper is assembled between the spacecraft bus and the attitude control actuators and acts to reduce the high frequency vibrations. The vibration isolation platform, which consists of the vibration isolation strut with magnetic suspension, has better performance in the region of high frequency according to the frequency domain analysis. An appropriate controller for the vibration isolation strut is designed based on the frequency domain analysis. Then, the attitude controller of the spacecraft bus is designed using the finite time control theory to reduce the low frequency vibrations, thus reducing the maneuvering time. Finally, the numerical simulations show that the vibration isolation platform and the attitude controller do work and cooperate well.

Arranging objects from a random and scattered distribution into an integral part has many applications, including bin packing, logistics, and other industrial fields. Measurement noises, manipulation uncertainties, models of irregular objects, and rich contacts bring considerable challenges to the improvements of the overall performance, such as seamlessness of the final pattern and task efficiency. In this paper, we propose an end-to-end reinforcement learning strategy that generates a series of pushing movements for scattered irregular objects inside a crate. An abstracted and sparse reward function is proposed to evaluate the pushing performance, and Proximal Policy Optimization (PPO) learning method that simultaneously trains a Convolutional Neural Network (CNN) and a fully connected actor network is developed for end-to-end decision making. The proposed method is evaluated in both simulation and real-world scenarios. The results show that the proposed method can arrange the scattered objects tightly to fit into each other in an efficient and flexible way, and can be transferred to the real world with unseen objects.